In a meta-analysis published in CMAJ, Ou and colleagues reported that, compared with use of conventional oxygen therapy, use of high-flow nasal cannula oxygen therapy could reduce the intubation rate in patients with acute hypoxemic respiratory failure.1 Although this study was well-designed, we identified several issues.
First, a search bias possibly existed, as 6 more trials were identified in our recent literature search (conducted September 2018).2–7 However, in our meta-analysis that included these additional trials (unpublished data, 2018),2–11 the conclusion remained consistent with that of Ou and colleagues.
Second, in our meta-analysis, the range of intubation rates was very large within these trials (from 0% in the study by Rittayamai and colleagues,7 to 42% in the study by Frat and colleagues8), indicating potential heterogeneity among these studies. Several well-designed studies8 and the trial by Hernández and colleagues9 also showed inconsistent findings.
Furthermore, we noticed that the severity of respiratory failure (Pao2:Fio2) at baseline differed among trials. Thus, we performed a subgroup meta-analysis to investigate whether the conclusion stayed consistent depending on Pao2:Fio2 levels. We divided the included trials into 3 subgroups according to the Pao2:Fio2 at enrolment (≥ 200, < 200 and unreported). In the subgroup with high Pao2:Fio2,3,6,9,10 use of high-flow nasal cannula oxygen therapy had a significant reduction in the intubation rate compared with use of conventional oxygen therapy. However, in the subgroup with low Pao2:Fio2,5,8 this benefit of high-flow nasal cannula oxygen therapy was nonsignificant.
Despite increased evidence supporting the use of high-flow nasal cannula oxygen therapy in acute respiratory failure, its inappropriate application may cause necessary intubation to be delayed, which is associated with poor outcomes.12 Thus, we suggest further studies are still needed to evaluate the benefit of high-flow nasal cannula oxygen therapy in patients with severe respiratory failure.
Footnotes
Competing interests: None declared.